Processing values of dead or retained biomass from timeseries output to fit the format required at the bottom of the forecast file. This can be used to map the catches resulting from forecasting with a particular harvest control rule into a model representing a different state of nature. This is a common task for US west coast groundfish but might be useful elsewhere.
Usage
SS_ForeCatch(
replist,
yrs = 2021:2032,
average = FALSE,
avg.yrs = 2016:2020,
total = NULL,
digits = 2,
dead = TRUE,
zeros = FALSE
)
Arguments
- replist
A list object created by
SS_output()
.- yrs
Range of years in which to fill in forecast catches from timeseries
- average
Use average catch over a range of years for forecast (as opposed to using forecast based on control rule)
- avg.yrs
Range of years to average over
- total
Either single value or vector of annual total forecast catch used to scale values (especially if values are from average catches). For west coast groundfish, total might be ACL for next 2 forecast years
- digits
Number of digits to round to in table
- dead
TRUE/FALSE switch to choose dead catch instead of retained catch.
- zeros
Include entries with zero catch (TRUE/FALSE)
Examples
if (FALSE) { # \dontrun{
# create table based on average over past 5 years
SS_ForeCatch(base, # object created by SS_output
yrs = 2021:2022, # years with fixed catch
average = TRUE, # catch by fleet from average catch
avg.yrs = 2014:2018
) # use average of catches over past 5 years
# create table with pre-defined totals where the first 2 years
# are based on current harvest specifications and the next 10 are set to some
# new value (with ratio among fleets based on average over past 5 years)
SS_ForeCatch(base, # object created by SS_output
yrs = 2021:2022, # years with fixed catch
average = TRUE, # catch by fleet from average catch
avg.yrs = 2016:2020, # use average of catches over past 5 years
total = c(rep(241.3, 2), rep(300, 10))
) # total
# create table based on harvest control rule projection in SS
# that can be mapped into an alternative state of nature
SS_ForeCatch(low_state, # object created by SS_output for low state
yrs = 2021:2032, # forecast period after fixed ACL years
average = FALSE
) # use values forecast in SS, not historic catch
} # }